Search Images Maps Play YouTube News Gmail Drive More »
Sign in
Screen reader users: click this link for accessible mode. Accessible mode has the same essential features but works better with your reader.

Patents

  1. Advanced Patent Search
Publication numberUS5521841 A
Publication typeGrant
Application numberUS 08/221,221
Publication dateMay 28, 1996
Filing dateMar 31, 1994
Priority dateMar 31, 1994
Fee statusPaid
Publication number08221221, 221221, US 5521841 A, US 5521841A, US-A-5521841, US5521841 A, US5521841A
InventorsFarshid Arman, Remi Depommier, Arding Hsu, Ming-Yee Chiu
Original AssigneeSiemens Corporate Research, Inc.
Export CitationBiBTeX, EndNote, RefMan
External Links: USPTO, USPTO Assignment, Espacenet
Browsing contents of a given video sequence
US 5521841 A
Abstract
A method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within the sequence and/or to decide whether the contents of a video sequence are relevant to a user, the video sequence having been pre-processed to detect scene changes and to build Rframes, and for allowing a user to scroll through the Rframes and to stop at a selected Rframe for processing, comprises playing the video sequence from a point of the selected Rframe, detecting all Rframes having respective degrees of similarity to the Rframe selected by the user, and presenting the similar Rframes to the user in a size or scale representative of the degrees of similarity.
Images(7)
Previous page
Next page
Claims(9)
We claim:
1. A computer-implemented method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within said sequence and/or to decide whether said contents of a video sequence are relevant to a user, said video sequence having been pre-processed to detect scene changes and to build Rframes, an Rframe being a representation of a sequence of images, and for allowing a user to scroll through said Rframes in a given manner and to stop at a selected Rframe for processing, said method comprising:
(a) playing said video sequence from the beginning of a shot represented by said selected Rframe to the end of said shot;
(b) detecting all Rframes having respective degrees of similarity to said Rframe selected by said user; and
(c) presenting said similar Rframes to said user in a size or scale representative of said degrees of similarity.
2. A method for representing contents of a video sequence in accordance with claim 1, wherein step (b) is performed by evaluating Rframe shape properties represented by using a respective moment for each Rframe image.
3. A method for representing contents of a video sequence in accordance with claim 2, wherein said moment of said R frame image is defined as:
mpq =ΣΣxp yq (x,y)    (2)
where, in reference to an image f(x,y), m00 corresponds to a total mass of a frame and m20 correspond to moments of inertia about the x and y axes, and a moment invariant derived therefrom is defined as ##EQU9## and the first few moment invariants are defined as
φ.sub. =η.sub. 2 +η.sub. 02
φ2 =(η2002)2 +4η11 2 ( 4)
φ3 =(η30 -3η12)2 +(3η2103)2 
The shape of each Rframe is then represented using the vector σ defined as:
σ={φ1, φ2, φ3, . . . , φ7 }(5)
and a "metric distance" ψ(α,β) is used to measure the similarity of two Rframes:
ψ(α,β)=|σ.sub.α -σ.sub.β |2 
4. A method for representing contents of a video sequence in accordance with claim 1, wherein step (b) is performed by evaluating Rframe color properties represented by using color histograms.
5. A method for representing contents of a video sequence in accordance with claim 1, wherein step (b) is performed by evaluating shape properties represented by using moments and Rframe color properties represented using color histograms.
6. A method for representing contents of a video sequence in accordance with claim 5, wherein said moment of said R frame is defined as:
mpq =ΣΣxp yq (x,y)    (2)
and a moment invariant derived therefrom is defined as ##EQU10## the first few moment invariants are defined as
φ.sub. =η.sub. 2 +η.sub. 02
φ2 =(η2002)2 +4η11 2 ( 4)
φ3 =(η30 -3η12)2 +(3η2103)2 
The shape of each Rframe is then represented using the vector σ defined as:
σ={φ1, φ2, φ3, . . . , φ7 }(5)
and a "metric distance" is used to measure the similarity of two Rframes:
ψ(α,β)=|σ.sub.α -σ.sub.β |2 
and
wherein a color histogram is obtained by counting how many times each color appears in each Rframe, using hue and saturation components of the HSI color space to calculate the color histogram for each Rframe; and
measuring similarity by the technique of histogram intersection, defined as: ##EQU11## where α and β are the two histograms, whereby the result of this intersection indicates how many pixels in one image have corresponding pixels of the same color in the other image, and the measure is normalized using: ##EQU12## where βi is the ith histogram.
7. A computer-implemented method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within said sequence and/or to decide whether said contents of a video sequence are relevant to a user, said video sequence having been pre-processed to detect scene changes and to build Rframes, an Rframe being a representation of a sequence of images, and for allowing a user to scroll in a given order through said Rframes and to stop at a selected Rframe for processing, said method comprising:
(a) playing said video sequence from the beginning of a shot represented by said selected Rframe to the end of said shot;
(b) detecting all Rframes having respective degrees of similarity to said Rframe selected by said user; and
(c) presenting Rframes having one of a predetermined degree of similarity to said Rframe selected by said user and predetermined degree of dissimilarity to said Rframe selected by said user.
8. A method for representing contents of a video sequence in accordance with claim 7, comprising the steps of:
comparing remaining, non-selected Rframes in said video sequence using moment and color histogram-based descriptions;
mapping similarity measures using: ##EQU13## where ζ=ε(α, βi) for mapping of color histogram intersection output of the equation
Ωhistogram [ε(α,βi)] ε {1,2,3}
and ζ=(α,βi) for mapping moment distance measure of the equation
Ωmoment [(α,βi)] ε {1,2,3}, where
Ω=3signifies very similar, Ω=2 somewhat similar, and Ω=1 not similar.
9. A method for representing contents of a video sequence in accordance with claim 8, wherein
if Ωmoment=1 then the output of said comparing using said color-histogram-based description is ignored;
if Ωmoment=3 then a final output is very similar, an only exception being when color-based output Whistogram=1 in which case said final output will also be 2, or somewhat similar; and
utilizing mapping from using said color-histogram-based description when Wmoment is not conclusive.
Description

The present invention relates to the task of browsing through video sequences. More particularly, the invention also relates to systems incorporating encoded video, wherein the ability to manage video data and display information efficiently is of particular importance and to low level management techniques for digital video.

For systems incorporating encoded video, such as video editing systems, various multimedia authoring systems, video-based training systems, and video on demand systems, the ability to manage video data and display information efficiently is critical. While known systems may incorporate other types of media as well, management of video is particularly difficult because of the vast volume of data associated with it and the high data rates involved, typically, many megabytes of data per minute. Prior steps taken towards the solution of video management problems have either relied on labor intensive techniques, such as manually entering keywords to describe the video contents, or on simple image processing techniques, such as analyzing histograms. These approaches have drawbacks and are neither close to providing ideal solutions, nor are they efficient in their tasks. Keywords have many drawbacks, such as, typically, an inadequate choice of terms for use at search time, the variable context in which the words are used, and the influence of the individual operator. See, for example, S-K. Chang and A. Hsu, Image information systems: Where do we go from here? IEEE Transactions on Knowledge and Data Engineering, 4(5):431-442, October 1992.

Furthermore, image processing steps cannot be efficiently applied to the hundreds of thousands of images that are usually associated with video. This paper presents techniques aimed at the management of encoded video, such as MPEG (D. Le Gall. MPEG: A video compression standard for multimedia applications, Communications of ACM, 34(4):46-58, April 1991.), JPEG (G. K. Wallace. The JPEG still picture compression standard, Communications of ACM, 34(4):30-44, April 1991.), and H.261 (M. Liou. Overview of the 64 kbits/s video coding standard, Communications of ACM, 34(4):59-63, April 1991.) which overcome the limitations of traditional image processing steps while enhancing keyword based approaches currently in wide use.

Sub-tasks of video management include the ability to quickly locate a particular video sequence--herein referred to as high level video management--and the ability to view particular points of interest within the video sequence--herein referred to as low level video management. The need for management of video exists in many domains, from TV news organizations where these capabilities are critical, to home video libraries where such capabilities can be very useful.

The present invention is concerned more particularly with low level management techniques for digital video. Currently, a widely used search technique, applicable, for example, to a tape recording machine, is to fast-forward and rewind to arrive at the point of interest. This technique is slow and inefficient. More recently, image processing techniques have been developed to operate on digital video in order to facilitate this task. A first step in solving this problem is to "divide" the video sequence into meaningful segments much like text in a book can be divided up into sentences. In video, a logical point to partition the video sequence is where the contents of video "change" in some way from one frame to the next--referred to as a scene change.

The past research work involving low level video management has concentrated on the parsing of video sequences into video clips. In most cases, the logical parsing point is a change in the camera view point or a change in the scene. Usually, the histogram of each scene is generated and a large change in the histogram from one scene to the next is used as a cutting point [11]. Ueda et.al suggest the use of the rate of change of the histogram instead of the absolute change to increase the reliability of the cut separation mechanism. H. Ueda, T. Miyatake, S. Sumino and A. Nagasaka, Automatic Structure Visualization for Video Editing, in InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr. 1993, pp. 137-141. Ueda et. al also consider the zooming and the panning of the camera; each video frame is divided into a number of non-overlapping small regions and in each region the optical flow of pixels belonging to that region is approximated and classified into zooming and panning of camera. This information is then stored along with each cut. Nagasaka and Tanaka studied various measures to detect the scene changes. A. Nagasaka and Y. Tanaka, Automatic video indexing and full-video search for object appearances. In E. Knuth and L. M. Wegner, editors, Proceedings of the IFIP TC2/WG2.6 Second Working Conference on Visual Database Systems, pages 113-127. North-Holland, Sep. 30-Oct. 3, 1991. The best measure according to their studies is a normalized c2 test to compare the distance between two histograms. Additionally, to minimize the effects of camera flashes and certain other noises, the frames are each divided into several subframes. Then, rather than comparing pairs of frames, every pair of subframes between the two frames are compared, the largest differences are discarded, and the decision is based upon the differences of the remaining subframes.

The use of DCT coefficients prior to decompression has been attempted previously in other applications. Hsu et. al use DCT compressed images in a military target classification system to discriminate between man-made and natural objects. Y. Hsu, S. Prum, J. H. Kagel, and H. C. Andrews, Pattern recognition experiments in mandala/cosine domain, IEEE Transactions on Pattern Analysis and Machine Intelligence, 5(5):512-520, September 1983. The Bhattacharyya distance discriminator is used to measure and rank numerous statistical calculations derived from the DCT coefficients; and it is in turn used in the decision making process. Smith and Rowe extended many properties of the cosine/Fourier transform to used the DCT coefficients to perform several algebraic operations on a pair of images. B. C. Smith and L. A. Rowe, Algorithms for manipulating compressed images. To appear in IEEE Computer Graphics and Applications, 13(5), September 1993. Scalar addition, scalar multiplication, pixel-wise addition, and pixel-wise multiplication operations on two images were defined using the DCT coefficients; these operations are used in video editing systems to perform such tasks as dissolving and subtitling.

Tonomura et. al introduced several approaches to view the contents of videoshots: variable speed, sampling flash, rush, and time-space browser. Y. Tonomura, A. Akutsu, K. Otsuji and T. Sadakata, VideoMAP and VideoSpaceIcon: Tools for Anatomizing Video Content, in InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr. 1993, pp. 131-136. Tonomura, Y. and Abe, S., Content Oriented Visual Interface Using Video Icons for Visual Database Systems, in Journal of Visual Languages and Computing, Vol. 1, 1990, pp. 183-198. The variable speed browser, is very similar to VCR's jog and shuttle functions; the sampling flash browser is a series of icons formed from the first frame of each video shot without any clues to the contents; in the rushbrowser, instead of using video shots the sequence is divided along equally spaced time intervals; and the time-space browser displays a temporal sequence on several icons. In Y. Tonomura, A. Akutsu, K. Otsuji and T. Sadakata, VideoMAP and VideoSpaceIcon: Tools for Anatomizing Video Content, in InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr. 1993, pp. 131-136, much emphasis is placed on characterizing the contents of video shots with respect to camera and object motions.

Similar to Tonomura, Elliot introduced a browser which stacks every frame of the sequence. This approach suffers from several shortcomings: First, the stack is built as the user is watching the sequence. E. Elliott, Watch, Grab, Arrange, See: Thinking With Motion Images via Streams and Collages, Ph.D. Thesis, MIT, February 1993. This is not useful for video browsing because the user is "forced" to watch the video sequence because the stack can make sense only once the video has been seen. The second shortcoming is that the stack holds only about 20 seconds of video; this amount of video is not practical for use in actual cases. Third, once the stack is built, the user may "stroke" the stack to watch the contents. This is a minor improvement, from the user's point of view, over FF/REW. This approach fails to provide the user with a basic browsing unit, and it is more appropriate for video editing than for browsing.

Zhang et. al used the video shot as their basic browsing unit. H-J. Zhang and W. Smoliar, Developing Power Tools for Video Indexing and Retrieval, in Proceedings of SPIE Conference on Storage and Retrieval for Image and Video Databases, San Jose, Calif., 1994. Similar to Tonomura, the frames of the shot are stacked to relay motion information and duration of the shot, and a frame from a shot may be "picked up" by placing the mouse along the side of the icon. In another mode, rather than stacking the frames, the icon thickness is used to convey shot duration; this is a wasteful use of screen space since the importance of the information does not justify the amount of screen space that is used.

Mills et. al introduced a browser for quick time video sequences. M. Mills, J. Cohen and Y-Y. Wong, A Magnifier Tool for Video Data, in Proceedings of ACM Computer Human Interface (CHI), May 3-7, 1992. Similar to Tonomura's rush browser, this browser does not take into consideration the contents of the video and rather systematically divides the sequence into several equal segments. Once the user has chosen a segment it in turn is divided into equal lengths and so on until the user can view each frame. In each case, the segment is represented using its first frame. This approach is a minor improvement to FF/REW and fails to provide the user with a sense of the contents of the video. The user could easily miss the information he or she is interested in because the representation of each segment has no relation to the reminder of the frames in that segment.

Disadvantages found in the foregoing above work are that either no basic browsing unit is used and/or that each frame of the video is needed by the user during the browsing operations, making it unsuitable for use over the network. Additionally, none of the above systems address the problem of icon management. This is very important since as many as several thousand icons could be needed to represent the shots for each two hour video sequence. Ueda et. al do address this issue by using color information. H. Ueda, T. Miyatake, S. Sumino and A. Nagasaka, Automatic Structure Visualization for Video Editing, in InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr. 1993, pp. 137-141. Color, however, cannot be the sole means of representation because color histograms are a many to one mapping functions. In our video browser, shape, as well as color information is used to help the user manage icons and navigate throughout a given video sequence.

In accordance with an aspect of the invention, a computer implemented method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within the sequence and/or to decide whether the contents of a video sequence are relevant to a user, the video sequence having been pre-processed to detect scene changes and to build Rframes, and for allowing a user to scroll through the Rframes in a given manner and to stop at a selected Rframe for processing, comprises (a) playing the video sequence from the beginning of a shot represented by the selected Rframe to the end of the shot; (b) detecting all Rframes having respective degrees of similarity to the Rframe selected by the user; and (c) presenting the similar Rframes to the user in a size or scale representative of the degrees of similarity.

In accordance with another aspect of the invention, step (b) is performed by evaluating shape properties represented by using a respective moment for each Rframe image.

In accordance with another aspect of the invention, step (b) is performed by evaluating color properties represented by using color histograms.

In accordance with another aspect of the invention, step (b) is performed by evaluating shape properties represented by using moments and color properties represented using color histograms.

In accordance with an aspect of the invention, a computer implemented method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within the sequence and/or to decide whether the contents of a video sequence are relevant to a user, the video sequence having been pre-processed to detect scene changes and to build Rframes, and for allowing a user to scroll in a given order through the Rframes and to stop at a selected Rframe for processing, the method comprises (a) playing the video sequence from the beginning of a shot represented by the selected Rframe to the end of the shot; (b) detecting all Rframes having respective degrees of similarity to the Rframe selected by the user; (c) storing said Rframes in a storage device for retrieval, as in a computer and (d) presenting Rframes having one of a predetermined degree of similarity to the Rframe selected by the user and predetermined degree of dissimilarity to the Rframe selected by the user.

In accordance with another aspect of the invention, if Wmoment=1 then the output of the comparing using the color-histogram-based description is ignored; if Wmoment=3 then a final output is very similar, an only exception being when color-based output Whistogram=1 in which case the final output will also be 2, or somewhat similar; and utilizing mapping from using the color-histogram-based description when Wmoment is not conclusive.

In accordance with an aspect of the invention, a computer implemented method for representing contents of a video sequence for allowing a user to rapidly view a video sequence in order to find a particular desired point within the sequence and/or to decide whether the contents of a video sequence are relevant to a user, the video sequence having been pre-processed to detect scene changes and to build Rframes, and for allowing a user to scroll either chronologically, based on degree of similarity, or in some other desired order, through the Rframes and to stop at a selected Rframe for processing, comprises playing the video sequence from a point of the selected Rframe, detecting all Rframes having respective degrees of similarity to the Rframe selected by the user, and presenting the similar Rframes to the user in a size or scale representative of the degrees of similarity.

The invention will be explained by way of exemplary embodiments and by reference to the drawing, helpful to an understanding of the invention, in which

FIG. 1 shows a representative frame, Rframe, for each video shot wherein are indicated (a) the structure of the Rframe, (b) motion tracking region; t=0 starts from the center of Rframe, (c)-(f) several examples;

FIG. 2 shows the browser in the basic mode of operation;

FIG. 3 shows the browser in the advanced mode of operation;

FIG. 4 shows the browser in the advanced mode of operation with prefs.;

FIG. 5 shows the frequency distribution (a) and block features (b) of DCT coefficients within a block; and

FIG. 6 shows an example of selecting subregions containing edges using the DCT coefficients. (a) The original frame. (b) The sub-regions found to contain no edges are shown in solid; the remaining regions may be decompressed for edge detection.

FIG. 7 shows an overview of the DCT and block concepts and the process of performing DCT transform on each frame of a video sequence.

Of significance to the present invention is a technique disclosed in a U.S. patent application entitled DETECTING SCENE CHANGES ON ENCODED VIDEO SEQUENCES, being filed concurrently herewith in names of Farshid Arman, Arding Hsu, and Ming-Yee Chiu and to a U.S. patent application entitled REPRESENTING CONTENTS OF SINGLE VIDEO SHOT USING RFRAMES and being filed concurrently herewith in the names of Farshid Arman, Adring Hsu, and Ming-Yee Chiu, both applications being under obligation of assignment to the same assignee as is the present application, and whereof the disclosures are herein incorporated by reference. Scene changes are readily detected using DCT coefficients in JPEG and MPEG encoded video sequences. See FIGS. 5 and 6. Within each 88 DCT block, the distribution of the DCT coefficients is used to classify the block as either type 0 (contains no high frequency components) or type 1 (contains high frequency components) The changes in the distributions of 0's and 1's from one frame to next is captured using eigenvectors and used to represent scene changes. The frames in between two consecutive scene changes form a video shot. Video shots may be thought of as the building blocks of video sequences, and are used in browsing, as herein disclosed in greater detail, database indexing, or any other operations that essentially form an abstraction of the video. To visualize each video shot, the content must be abstracted in a meaningful manner such that it is representative of the contents of the shot; this is achieved using representative frames or Rframes, as herein disclosed in greater detail.

Of particular significance is the problem of detecting scene changes on encoded video sequences, particularly in the context of rapidly viewing the contents of a given video sequence, a process herein referred to as browsing. Browsing through video sequences is a critical requirement in many domains and applications in which the user is either required to choose a few video sequences from among many, and/or the user has to find a particular point within a single video sequence.

Such cases arise in numerous situations, such as in remote access of video, video database navigation, video editing, video-based education and training, and, in the near future, video e-mail and recorded desk-top video conferencing sessions. In such cases, the user must view the contents of the video sequences in order to choose the most relevant or to locate a desired point. Assigned labels, keyword descriptions, and database indexing may be useful in reducing the number of possibilities somewhat; however, in many cases the user is still left to decide among at least a number of possibilities. Consider, for instance, the case in which the user has submitted a query to a remote database and the database search has resulted in the offer of several possibilities. At this point the user must decide if the context and contents of the returned videos match the requirements. This may only be achieved by viewing each of the returned videos. Viewing video would require that each video be retrieved from, typically, a hierarchical storage system, transmitted over the network in its entirety as the user plays the video or, at most, fast forwards and rewinds. This process is time consuming, inefficient, not cost effective, and wasteful of bandwidth.

Abstractions of each of the video sequences are pre-computed and the abstractions are retrieved from the system, transmitted, as may be needed, and viewed by a user. The abstractions are many orders of magnitude smaller in size than the video sequences themselves, and thus, the system's response time, bandwidth needs, and, most importantly, the user's viewing time are reduced. In addition, the proposed system allows the user to rapidly pinpoint a desired location within a video sequence.

In accordance with an aspect of the invention, content-based video browsing is achieved by pre-processing steps which are performed off-line before the user gains access:

(a) detect scene changes in the compressed video to form video shots; and

(b) construct the abstractions for each video shot to represent the contents.

The abstractions are referred to as Rframes. Additionally, a number of steps are performed during browsing which are driven by the users' particular needs:

(c) present the Rframes so that the user can easily search the contents of the video sequence; and

(d) apply a technique to manage the Rframes comprising combining similarity measurements based on shape and color.

Processing during the browsing is necessary because each user may be different and may have varying needs at different times even for the same sequence.

In accordance with the present invention, the methodology herein disclosed represents the contents of a video sequence. The representation is used to allow the user to rapidly view a video sequence in order to find a particular point within the sequence and/or to decide whether the contents of the sequence are relevant to his or her needs. This system, referred to as content-based browsing, forms an abstraction, as herein disclosed in greater detail, to represent each detected shot, of the sequence by using a representative frame, or an Rframe, as herein disclosed, and it includes management techniques to allow the user to easily navigate the Rframes. This methodology is superior to the current techniques of fast forward and rewind because rather than using every frame to view and judge the contents, only a few abstractions are used. Therefore, the need to retrieve the video from a storage system and to transmit every frame over the network in its entirety no longer exists, saving time, expenses, and bandwidth.

Content-based browsing is advantageous over the fast forward and rewind technique (FF/REW) while nevertheless being as convenient to use. Using FF/REW the user must view every frame at rapid speeds, with the likelihood of missing shots that last a short period, while being forced to watch long lasting and possibly irrelevant shots. In addition, users searching for a specific point within a sequence are typically forced to refine their search after a number of fast forward and rewind operations until the video is at the precise point of interest, a time-consuming and tedious task. In the content-based browser in accordance with the invention, the exact points of scene changes are defined internally, and no "fine tuning" by the user is necessary. It is noteworthy that the above described disadvantages of FF/REW persist even on digital video and on other random access media, such as laser disks. Lastly, FF/REW as the means for browsing of digital video is extremely inefficient considering, the expense of accessing disks and/or tapes, decoding, and transmission.

In relation to the processing of compressed video efficiently for scene change detection, selective decoding is utilized to take advantage of the information already encoded in the compressed data; specifically a discrete cosine transform (DCT)-based standard such as JPEG (see G. K. Wallace, "The JPEG still picture compression standard", Communications of ACM, 34(4):30-44, April 1991) or H.261 (M. Liou. Overview of the 64 kbits/s video coding standard, Communications of ACM, 34(4):59-63, April 1991.) and many processing steps needed on every frame of a video sequence are performed prior to full decompression. The DCT coefficients are analyzed to systematically detect scene changes or video cuts which are used in browsing or in further feature extraction and indexing. In the past, expensive operations such as color histogram analysis, have been performed on every frame to achieve the same tasks. D. Le Gall. MPEG: A video compression standard for multimedia applications. Communications of ACM, 34(4):46-58, April 1991.

The encoding standards process begins with dividing each color component of the image into a set of 88 blocks. FIG. 7 shows an overview of the DCT and block concepts. The pixels in the blocks are then each transformed using the forward discrete cosine transform (DCT): ##EQU1## where C(τ)=1/(√2) if τ=0 and 1 otherwise, F(u,v) are the DCT coefficients, and f(x,y) are the input pixels. F(0,0) is the DC term--the average of the 64 pixel values, and the remaining 63 coefficients are termed the AC coefficients. The 64 coefficients from each block are then quantized to preserve only the visually significant information: ##EQU2## where Q(u,v) are the elements of the quantization table, and [ ] represents the integer rounding operation. The coefficients are then encoded in a zig-zag order by placing the low order frequency components before the high frequency components. The coefficients are then encoded using the Huffman entropy encoding. The processing presented next assumes that the encoded data has partially been decoded by applying the Huffman decoder and the resultant coefficients may or may not have been dequantized depending on the quantization table. See FIG. 5 for the frequency distribution (a) and block features (b) of DCT coefficients within a block. Zero coefficients in the "high" regions indicate that the 88 block has low frequency components only and substantially no high frequency components. See FIG. 6 for an example of selecting subregions containing edges using the DCT coefficients. (a) The original frame. (b) The sub-regions found to contain no edges are shown in solid; the remaining regions may be decompressed for edge detection.

The approach herein differs from previous solutions in that, inter alia, unlike prior methods wherein all steps are performed on decompressed video frames, the present approach takes advantage of the fact that the incoming video is already in the compressed form. Thereafter, the information that is already encoded in the compression process is utilized to take advantage of several facts: first, the computational cost of fully decompressing every frame is not necessary and is saved when only a selected number of frames are chosen prior to decompression for further processing or for browsing. Second, coefficients in the spatial frequency domain are mathematically related to the spatial domain, and they may directly be used in detecting changes in the video sequence. Third, the knowledge of the blocks' location preserves spatial domain information to a certain extent.

The scene change detection is performed by the application of a programmed computer in accordance with the following method or "algorithm":

(a) examine each DCT block in the compressed video frame, and if high frequency coefficients exist mark that block as 1, else mark that block as 0. The output of this step is a matrix of 0s and 1s. The size of this matrix is determined by the size of the video frame divided by 8 length wise and width wise. For example, a 320240 video frame will yield a 4030 matrix;

(b) delete columns or rows to transform the matrix of step 1 into a square matrix; for example delete 10 columns to obtain a 3030 matrix. Preferably, for every frame of the video, the same corresponding columns or rows are deleted. This step may include subsampling to reduce the matrix size. For example, delete every other row and column. The final output from this step is an nn matrix;

(c) derive the two principal vectors of the matrix, to describe the contents of each video frame, in accordance with principles of linear algebra that state that each nn matrix has at least one and at most n eigenvalues: λi, 1≧i≧n. and for two dimensional shapes there will be 2 eigenvalues, that each eigenvalue will have a corresponding eigenvector, and that these two vectors are the principal vectors of the matrix;

(d) detect a change in the content from one video frame to the next, or scene changes, by utilizing the inner product to detect such change, since a change in the content from one video frame to the next, or scene changes, will also cause the vectors to change in accordance with the following expression: ##EQU3## where Δ is the temporal distance in between two frames; and

(e) if, ∂, 1≧∂≧0, is larger than a threshold, τ, then indicate that a scene change has occurred.

The video content in between two scene changes is labeled as a "shot".

If the format of the video is motion JPEG, then the DCT coefficients of step (a) are obtained from each frame and Equation (1) is applied as stated in step (e). In case the format is MPEG where three types of frames are defined (I, B, and P), each two frames in Equation (1) must be of the same type; i.e, Equation (1) cannot compare an I frame with the neighboring B or P frame.

Reference is made to the afore-mentioned U.S. patent application entitled REPRESENTING CONTENTS OF SINGLE VIDEO SHOT USING RFRAMES and being filed concurrently herewith in the names of Farshid Arman, Adring Hsu, and Ming-Yee Chiu and being subject to an obligation of assignment to the same assignee as is the present application. Each detected shot is represented using an Rframe, which is designed to allow the user to perform five tasks: first, to be able to judge the contents of the shot. Second, to decide if the scene change detection may have missed a shot. While many of the proposed scene change detectors have high accuracy rates of 90% and above, none claims 100% accuracy; in addition, many complicated transitions can cause false negatives during scene change detection. Therefore, from the user's point of view, it is desirable that there be a mechanism to ensure the user that no scene changes have been missed during this shot. The third task of the Rframe is to provide the user with the sense of motion within the shot. The fourth feature allows the user to easily determine the length or duration of the shot in seconds. The fifth allows the user to determine if any captions appear in the video shot. In order to form the Rframes the video sequence must have already been divided into meaningful segments, such as video shot (the frames in between two consecutive scene changes form a video shot, as herein disclosed. The collection of Rframes is used to represent the contents of the entire video sequence in browsing and in navigation operations, as herein explained in relation to browsing the contents of a given video sequence.

Each Rframe comprises a body, four motion tracking regions, shot length indicators and a caption indicator. See FIG. 1. The body of the Rframe is a frame chosen from the video shot; currently, the tenth frame is chosen, but other possibilities exist, such as the last frame for zoom-in shots. The motion tracking regions trace the motion of boundary pixels through time; hence they can be used as guides to camera, or global, motion. The motion tracking regions also serve as an indicator of missed scene changes. In case the shot contains a scene change, the tracking of boundary pixels will "fail" causing a straight line to appear in the motion tracking region (see FIG. 1-e). The time indicators are designed so that a brief glance at each Rframe allows the user to determine if the corresponding shot is long or short while a more precise estimation of the length of the shot is also possible well by counting the 2 and 4 second squares. This representation of shot length does not occupy any valuable screen space; printing the exact number of seconds on the other hand would not allow the user to quickly compare shot lengths.

In FIG. 1, a representative frame, Rframe, for each video shot is shown. (a) shows the structure of the Rframe, (b) shows motion tracking region; t=0 starts from the center of Rframe, (c)-(f) show several examples: (c) the anchorman has moved his hands but the camera is stationary as is evidenced by the straight lines, and the shot contains a caption; (d) shows that the camera has panned to the left following the motion of the animal, the curves start (t=0) and move to the right, no captions are present in this shot; (e) shows an example of a missed scene change, the straight lines not in contact with the center indicate the possibility that the shot may contain a scene change; (f) shows that the camera is stationary but the objects have moved in various directions; this shot contains a caption.

To construct the motion tracking regions, the shot is sub-sampled to select a few of the frames. Four slices, one from each side, of each selected frame are then stacked and an edge detection algorithm is applied to each of the four stacks. This simple operation in effect tracks the border pixels from one frame to the next enabling the user to visualize the motion.

Edge detection is a local operation performed using the principles of convolution. A mask which is an mm matrix is convolved with the pixels ineach of the motion tracking regions. The output of the convolution highlights the pixels where there are changes in two neighboring pixels, where neighboring means left, right, top, or bottom. Many mm matrices exist, such as the Laplacian matrix: ##EQU4##

Reference is made to Gonzalez, op. cit. for more details.

As mentioned earlier, video sequences require a "basic browsing unit" which can be used in browsing, and unlike the case of newspapers or books where an editor manually chooses the headline for each article or chapter, the process of choosing the video browsing unit must be automatic. This is because of the vast amount of data that will exist in the video sequences. Furthermore, manual intervention would inherently incorporate extrinsic influences into the material. This influence could in turn impede a user's search by providing false leads or not enough leads and thereby requiring the user to use FF/REW. While the process of choosing the video browsing unit must be automatic, its result must also be meaningful to the user because this is the tool used to decide whether the returned video sequences are relevant to the task at hand. A remaining issue in designing a videobrowser is its speed; the video browser must be significantly faster, as compared with FF/REW, while remaining convenient to use.

A video browser disclosed herein in accordance with the present invention satisfies such requirements. The present video browser uses shots as the basic building blocks of a video sequence characterized by the use of "representative frames", or Rframes. The sequences in the video collection are pre-processed once to detect the scene changes and to build the Rframes. Then, to browse (through) a particular videosequence, the user may scroll through all the Rframes to view the visual contents of the sequence. Once the user has chosen an Rframe, the corresponding video shot may be played back. Further information, such as the length of each shot and the approximate motions, are readily represented as well. In cases in which several hundred scenes, and therefore several hundred Rframes, may exist in a given video sequence, advanced techniques are used to allow the user to easily manage the information.

At start up, the browser displays the precomputed Rframes in chronological order, (see FIG. 2, which shows the browser in the basic mode of operation. The row of Rframes is on the bottom, and the sequence at the point chosen by the user is displayed on top. The user may play the video from that pointand automatically stop at the end of the shot, or continue past the scene change.). The user may scroll through the Rframes and once an Rframe is chosen, then the video is played from precisely that point. The user's second option is to choose one Rframe and view all other similar Rframes. The degree to which each Rframe in the sequence is similar to the chosen Rframe is conveyed to the user by varying the size of each Rframe. The most similar Rframes are displayed at their original scale, somewhat similar Rframes are displayed at a smaller scale, for example, at a default value of 33% scale, and the dissimilar Rframes are displayed at even a smaller scale (default 5%), see FIG. 3, which shows the browser in the advanced mode of operation. The top rowis the original set of Rframes, the user has the chosen one Rframe (outlined by the red square) and the bottom row show all other similar Rframes, somewhat similar Rframes are shown at 33% of the originalwidth, and non-similar Rframes are shown at 5% of the original width-scene as black bars. The defaults are easily adjustable by the user (see FIG. 4, which shows the browser in the advanced mode of operation with prefs.). The browser in the advanced mode of operation as the user is choosing how to view each grouping category in the preferences window. The shown setting indicates that the somewhat and not similar Rframes be shown as black bars, and only the similar Rframes are shown at full scale.).

In addition to asking similar Rframes to be displayed on the second row of the browser, the user can combine several requests: show Rframes that are "similar" to Rframe X and "not similar" to Rframe Y. After each request the results are shown on a new reel of Rframes. Therefore, the user may have several reels at any time each containing a different "view" of the same sequence. The user's requests may be performed on any one of reels and the results displayed in a new reel or by overwriting an existing one depending on user's preferences.

As mentioned earlier, the browser must be as convenient to use as the current method of FF/REW. The proposed browser satisfies this criterion; the only user required actions are scroll and single or double clicks on a control mouse.

Assuming the scene changes have been detected, several issues arise when there are numerous Rframes--for example, more than the user can easily search and navigate through. As mentioned earlier, the user may choose one Rframe and ask the system to return all similar Rframes in the same videosequence. The key to measure this similarity effectively and correctly is the means by which each Rframe is represented internally. Representations are used to describe Rframes, a key issue in the field of computer vision. The representations dictate the matching strategy, its robustness, and the system's efficiency. Also, the descriptions are used in the calculations of various properties of objects in the scene needed during the grouping stage. In almost all cases, the two-dimensional array of numbers used to display the Rframes is not very useful in its "raw" form.

The browser uses two representation schemes which complement one another: Shape properties represented using moments, and color properties represented using color histograms. Both representation schemes are insensitive to minor changes in the scene, such as object motion, viewing distance, and so forth, and both are compact representations allowing for efficient similarity measurements. The following two sections describe these representation schemes and their usage in more detail.

The shape of objects within an Rframe is the main property used in Rframe management, and it is represented using moment invariants. The moment of an image f(x,y) is defined as:

mpq =ΣΣxp yq (x,y)    (2)

A physical interpretation of moments is possible if the grey level of each Rframe is regarded as its mass; then, in such an analogy, m00 would be the total mass of an Rframe and m20 and m02 would be the moments of inertia around the x and y axes. Moments invariants exhibit characteristics which makes them an ideal representation mechanism in the video browser. Invariance with respect to any of scale change, rotation, and translation, are some of such characteristics which are used in the browser to describe Rframes. Moment invariants are derived from normalized central moments defined as: ##EQU5##

Then, the first few moment invariants are defined as (M.-K. Hu, Pattern Recognition by moment invariants, in Proc. IRE, Vol. 49, 1961, p. 1428. M.-K. Hu, Visual pattern recognition by moment invariants, in IRE Trans. Inform. Theory, Vol. 8, February 1962, pp. 179-187. R. Gonzalez and P. Witz, Digital Image Processing, Addison-Wesley, Readings, Mass., 1977.):

φ12002 

φ2 =(η2002)2 +4η11 2 (4)

φ3 =(η30 -3η12)2 +(3η2103)2 

The shape of each Rframe is then represented using the vector defined as:

σ={φ1, φ2, φ3, . . . , φ7 }(5)

Finally, the Euclidean distance (this distance may be the dot product of the two vectors, so that in general, a "metric" distance can be measuring similarity) is used to measure the similarity of two Rframes:

ψ(α, β)=|σ.sub.α -σ.sub.β |2                                          (6)

Color is the second feature used extensively in Rframe management in accordance with the present invention. Color has many of the characteristics of moments, such as the ability to simply represent, or describe each Rframe. Contrary to the case of moments, however, it is less sensitive to differences, such as due to motion within a frame. Color cannot be the sole representation of Rframe contents because most means of representing color rely on color histograms which by definition are plurality-to-one mapping functions. Hence, many completely different Rframes, or video frames, may have very similar color representations. Color histograms alone are not sufficient to detect any differences in a red and white checkered board versus a white board with red parallel lines, for example, since the color contents of the two can be identical.

The browser represents the color contents of each Rframe using the color histogram, which is essentially the frequency distribution function of the color of each pixel. Given a color model (RGB, HSI, etc.), the histogram is obtained by counting how many times each color appears in each Rframe (see C. L. Novak and S. A. Shafer, Anatomy of a Color Histogram, in Proceeding of Computer Vision and Pattern Recognition, Champaign, Ill., June, 1992, pp. 599-605 for more details). It is herein recognized to use the hue and saturation components of the HSI color space, in accordance with the inventors' previous work: F. Arman, A. Hsu and M-Y. Chiu, Image Processing on Encoded Video Sequences, in ACM Multimedia Systems Journal, to appear 1994 to calculate the color histogram for each Rframe. In order to measure the similarity of two given Rframes, the technique of histogram intersection known from Swain and Ballard (Swain, M. J. and Ballard, D. H., Color Indexing, in Int. J. of Computer Vision, Vol. 7, No. 1, 1991, pp. 11-32) is herein applied. The intersection of two histograms is defined as: ##EQU6## where α and β are the two histograms. The result of this intersection indicates how many pixels in one image have corresponding pixels of the same color in the other image, and the measure is normalized using: ##EQU7## where βi is the ith histogram.

Once the user has chosen an Rframe, the moments and the color histogram of that Rframe are compared to the remaining Rframes. The output of the moment-based and color histogram-based analyses are two floating point numbers describing the similarity in shape and in color of the Rframes' body. In order to combine and compare these two different entities a mapping function is used which maps both entities onto a common space. This is performed using: ##EQU8## where ζ=ε(α, βi) for mapping of color histogram intersection output of Equation (7):

Ωhistogram [ε(α,βi)] ε {1,2,3}(10)

and ζ=(α,βi) for mapping moment distance measure of Equation (5):

Ωmoment [(α,βi)] ε {1,2,3}(11)

Ω=3 signifies very similar, Ω=2 somewhat similar, and Ω=1 not similar.

The rules of Table 1 are then used to combine the mapped properties. Generally, the output of

              TABLE 1______________________________________moment           color  final______________________________________3                3      33                2      33                1      22                3      32                2      22                1      11                3      11                2      11                1      1______________________________________ The rules for combining the results of the momentbased and histogrambased matching: 3 = very similar, 2 = somewhat similar, and 1 = not similar.

moments carries a bigger weight (see Table 1). If Wmoment=1 then the output of the color-histogram-based analysis is ignored; i.e., the final output will always be that the two Rframes under examination are not similar. If Wmoment=3 then the final output is also very similar, the only exception being when color-based output Whistogram=1 in which case the final output will also be 2, or somewhat similar. The mapping from color histogram is used when Wmoment is not conclusive; i.e., Wmoment=2; in this case the final output is set to the value of the color histogram mapping.

The processing time for the grouping takes advantage of two points. First, the moments and the histograms are calculated a priori and the only step needed at run time is measuring similarity; i.e., applying Equation (2) and Equation (8). Second, using the rules specified in Table 1, the histogram intersection operation, the more expensive of the two operations, has to be performed on a subset of the Rframes providing additional time saving steps. It is also contemplated within the context of the present invention to utilize an indexing scheme to store the histogram and the moment calculations; this greatly speeds up the grouping time.

Reference is also made to Tonomura, Y. and Abe, S., Content Oriented Visual Interface Using Video Icons for Visual Database Systems, in Journal of Visual Languages and Computing, Vol. 1, 1990, pp. 183-198.

It should be clearly understood that the foregoing embodiments are practiced by the use of a programmed digital computer. The invention has been explained by way of exemplary embodiments. However, it will be understood that various changes and modifications will be apparent to one of skill in the art to which the present invention pertains, but such changes and modifications are understood to be within the spirit of the invention whose scope is defined by the claims following.

Patent Citations
Cited PatentFiling datePublication dateApplicantTitle
US4685003 *Dec 2, 1983Aug 4, 1987Lex Computing & Management CorporationVideo composition method and apparatus for providing simultaneous inputting and sorting of video source material
US4755956 *Nov 1, 1985Jul 5, 1988Allied-Signal Inc.Freeze frame apparatus for moving map display system
US4876597 *Aug 18, 1988Oct 24, 1989Adt Security Systems, Inc.Video observation systems
US4943866 *Jun 16, 1989Jul 24, 1990Lex Computer And Management CorporationVideo composition method and apparatus employing smooth scrolling
US4949193 *Jun 16, 1989Aug 14, 1990Lex Computer And Management CorporationVideo composition method employing action scrolling
US4970663 *Apr 28, 1989Nov 13, 1990Avid Technology, Inc.Method and apparatus for manipulating digital video data
US4979050 *Dec 2, 1983Dec 18, 1990Lex Computer And Management CorporationVideo composition method for assembling video segments
US5065345 *Aug 21, 1990Nov 12, 1991Dyned International, Inc.Interactive audiovisual control mechanism
US5083860 *Mar 11, 1991Jan 28, 1992Institut For Personalized Information EnvironmentMethod for detecting change points in motion picture images
US5101364 *Feb 9, 1990Mar 31, 1992Massachusetts Institute Of TechnologyMethod and facility for dynamic video composition and viewing
US5109482 *Feb 19, 1991Apr 28, 1992David BohrmanInteractive video control system for displaying user-selectable clips
US5136395 *Jun 5, 1990Aug 4, 1992Pioneer Electronic CorporationStill image signal reproducing apparatus and method for operation
US5177513 *Mar 13, 1992Jan 5, 1993Kabushiki Kaisha ToshibaMoving picture managing device and method of managing a moving picture
US5179449 *Jan 11, 1990Jan 12, 1993Kabushiki Kaisha ToshibaScene boundary detecting apparatus
US5267351 *Dec 22, 1989Nov 30, 1993Avid Technology, Inc.Media storage and retrieval system
US5283864 *Oct 30, 1990Feb 1, 1994Wang Laboratories, Inc.In a data processing system
US5307456 *Jan 28, 1992Apr 26, 1994Sony Electronics, Inc.Integrated multi-media production and authoring system
US5317732 *Apr 26, 1991May 31, 1994Commodore Electronics LimitedSystem for relocating a multimedia presentation on a different platform by extracting a resource map in order to remap and relocate resources
US5355450 *Apr 10, 1992Oct 11, 1994Avid Technology, Inc.Media composer with adjustable source material compression
US5388197 *Mar 29, 1994Feb 7, 1995The Grass Valley Group, Inc.Video editing system operator inter-face for visualization and interactive control of video material
US5404316 *Aug 3, 1992Apr 4, 1995Spectra Group Ltd., Inc.Desktop digital video processing system
Non-Patent Citations
Reference
1"A Magnifier Tool for Video Data", Mills et al., Proceedings of ACM Computer Human Interface (CHI), May 3-7, 1992, pp. 93-98.
2"A New Family of Algorithms for Manipulating Compressed Images", Smith et al., IEEE Computer Graphics and Applications, 1993.
3"Anatomy of Color Histogram", Novak et al., Proceeding of Computer Vision and Pattern Recognition, Champaign, IL, Jun. 1992, pp. 599-605.
4"Automatic Structure Visualization for Video Editing", Ueda et al., InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr. 1993, pp. 137-141.
5"Automatic Video Indexing and Full-Video Search for Object Appearances", Nagasaka et al., Proceedings of the IFIP TC2/WG2.6 Second Working Conference on Visual Database Systems, North Holland, Sep. 30-Oct 3, 1991. pp. 113-127.
6"Color Indexing", Swain et al., International Journal of Computer Vision, vol. 7, No. 1, 1991, pp. 11-32.
7"Content Oriented Visual Interface Using Video Icons for Visual Database Systems", Tonomura et al., Journal of Visual Languages and Computing (1990) 1, pp. 183-198.
8"Developing Power Tools for Video Indexing and Retrieval", Zhang et al., Proceedings of SPIE Conference on Storage and Retrieval for Image and Video Databases, San Jose, CA, 1994.
9"Image Information Systems: Where Do We Go From Here?", Chang et al., IEEE Transactions on Knowledge and Data Engineering, vol. 4, No. 5, Oct. 1992, pp. 431-442.
10"Image Processing on Compressed Data for Large Video Databases", Arman et al., Proceedings of First ACM International Conference on Multimedia, Anaheim, CA, 1-6 Aug. 1993, pp. 267-272.
11"Image Processing on Encoded Video Sequences", Arman et al., ACM Multimedia Systems Journal, to appear 1994.
12"Impact: An Interactive Natural-Motion-Picture Dedicated Multimedia Authoring System", Ueda et al., Proceedings of Human Factors in Computing Systems (CHI 91), New Orleans, LA, Apr. 27-May 2, 1991, pp. 343-350.
13"MPEG: A Video Compression Standard for Multimedia Applications, Le Gall, Communications of the ACM", vol. 34, No. 4, Apr. 1991, pp. 47-58.
14"News On-Demand for Multimedia Networks", Miller et al., ACM International Conference on Multimedia, Anaheim, CA, 1-6 Aug. 1993, pp. 383-392.
15"Overview of the px64 kbit/s Video Coding Standard", Liou, Communications of the ACM, vol. 34, No. 4, Apr. 1991, pp. 60-63.
16"Pattern Recognition by Moment Invariants", Hu et al., Proc. IRE, vol. 49, 1961, p. 1428.
17"Pattern Recognition Experiments in the Mandala/Cosine Domain", Hsu et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-5, No. 5, Sep. 1983, pp. 512-520.
18"The JPEG Still Picture Compression Standard", Wallace Communications of the ACM, vol. 34, No. 4, Apr. 1991, pp. 31-44.
19"The Revised Fundamental Theorem of Moment Invariants", Reiss, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 13, No. 8, Aug. 1991, pp. 830-834.
20"VideoMAP and VideoSpaceIcon: Tools for Anatomizing Video Content", Tonomura et al., InterCHI'93 Conference Proceedings, Amsterdam, The Netherlands, 24-29 Apr., 1993, pp. 131-136.
21"Visual Pattern Recognition by MomentInvariants", IRE Trans. Inform. Theory, vol. 8, Feb. 1962, pp. 179-187.
22"Watch-Grab-Arrange-See: Thinking with Motion Images via Streams and Collages", Elliott, Ph.D. Thesis, MIT, Feb. 1993.
23 *A Magnifier Tool for Video Data , Mills et al., Proceedings of ACM Computer Human Interface (CHI), May 3 7, 1992, pp. 93 98.
24 *A New Family of Algorithms for Manipulating Compressed Images , Smith et al., IEEE Computer Graphics and Applications, 1993.
25 *Anatomy of Color Histogram , Novak et al., Proceeding of Computer Vision and Pattern Recognition, Champaign, IL, Jun. 1992, pp. 599 605.
26Arman et al., "Feature Management for Large Video Databases", 1993. (Abstract Only).
27 *Arman et al., Feature Management for Large Video Databases , 1993. (Abstract Only).
28 *Automatic Structure Visualization for Video Editing , Ueda et al., InterCHI 93 Conference Proceedings, Amsterdam, The Netherlands, 24 29 Apr. 1993, pp. 137 141.
29 *Automatic Video Indexing and Full Video Search for Object Appearances , Nagasaka et al., Proceedings of the IFIP TC2/WG2.6 Second Working Conference on Visual Database Systems, North Holland, Sep. 30 Oct 3, 1991. pp. 113 127.
30 *Color Indexing , Swain et al., International Journal of Computer Vision, vol. 7, No. 1, 1991, pp. 11 32.
31 *Content Oriented Visual Interface Using Video Icons for Visual Database Systems , Tonomura et al., Journal of Visual Languages and Computing (1990) 1, pp. 183 198.
32 *Developing Power Tools for Video Indexing and Retrieval , Zhang et al., Proceedings of SPIE Conference on Storage and Retrieval for Image and Video Databases, San Jose, CA, 1994.
33Iino et al., "An Object-Oriented Model for Spatio-Temporal Synchronization of Multimedia Information", May, 1994.
34 *Iino et al., An Object Oriented Model for Spatio Temporal Synchronization of Multimedia Information , May, 1994.
35 *Image Information Systems: Where Do We Go From Here , Chang et al., IEEE Transactions on Knowledge and Data Engineering, vol. 4, No. 5, Oct. 1992, pp. 431 442.
36 *Image Processing on Compressed Data for Large Video Databases , Arman et al., Proceedings of First ACM International Conference on Multimedia, Anaheim, CA, 1 6 Aug. 1993, pp. 267 272.
37 *Image Processing on Encoded Video Sequences , Arman et al., ACM Multimedia Systems Journal, to appear 1994.
38 *Impact: An Interactive Natural Motion Picture Dedicated Multimedia Authoring System , Ueda et al., Proceedings of Human Factors in Computing Systems (CHI 91), New Orleans, LA, Apr. 27 May 2, 1991, pp. 343 350.
39Iyengar et al., "Codes Designs for Image Browsing", 1994.
40 *Iyengar et al., Codes Designs for Image Browsing , 1994.
41Mackay et al., "Virtual Video Editing in Interactive Multimedia Applications", 1989.
42 *Mackay et al., Virtual Video Editing in Interactive Multimedia Applications , 1989.
43Mills, "Media Composition for Casual Users", 1992.
44 *Mills, Media Composition for Casual Users , 1992.
45 *MPEG: A Video Compression Standard for Multimedia Applications, Le Gall, Communications of the ACM , vol. 34, No. 4, Apr. 1991, pp. 47 58.
46 *News On Demand for Multimedia Networks , Miller et al., ACM International Conference on Multimedia, Anaheim, CA, 1 6 Aug. 1993, pp. 383 392.
47 *Overview of the px64 kbit/s Video Coding Standard , Liou, Communications of the ACM, vol. 34, No. 4, Apr. 1991, pp. 60 63.
48 *Pattern Recognition by Moment Invariants , Hu et al., Proc. IRE, vol. 49, 1961, p. 1428.
49 *Pattern Recognition Experiments in the Mandala/Cosine Domain , Hsu et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI 5, No. 5, Sep. 1983, pp. 512 520.
50Rangan et al., "A Window-based Editor for Digital Video and Audio", Jan. 1992.
51 *Rangan et al., A Window based Editor for Digital Video and Audio , Jan. 1992.
52 *The JPEG Still Picture Compression Standard , Wallace Communications of the ACM, vol. 34, No. 4, Apr. 1991, pp. 31 44.
53 *The Revised Fundamental Theorem of Moment Invariants , Reiss, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 13, No. 8, Aug. 1991, pp. 830 834.
54 *VideoMAP and VideoSpaceIcon: Tools for Anatomizing Video Content , Tonomura et al., InterCHI 93 Conference Proceedings, Amsterdam, The Netherlands, 24 29 Apr., 1993, pp. 131 136.
55 *Visual Pattern Recognition by MomentInvariants , IRE Trans. Inform. Theory, vol. 8, Feb. 1962, pp. 179 187.
56 *Watch Grab Arrange See: Thinking with Motion Images via Streams and Collages , Elliott, Ph.D. Thesis, MIT, Feb. 1993.
Referenced by
Citing PatentFiling datePublication dateApplicantTitle
US5659539 *Jul 14, 1995Aug 19, 1997Oracle CorporationMethod and apparatus for frame accurate access of digital audio-visual information
US5664227 *Oct 14, 1994Sep 2, 1997Carnegie Mellon UniversitySystem and method for skimming digital audio/video data
US5682326 *Apr 3, 1995Oct 28, 1997Radius Inc.Desktop digital video processing system
US5694170 *Apr 6, 1995Dec 2, 1997International Business Machines CorporationVideo compression using multiple computing agents
US5713021 *Sep 14, 1995Jan 27, 1998Fujitsu LimitedMultimedia data search system that searches for a portion of multimedia data using objects corresponding to the portion of multimedia data
US5732146 *Apr 18, 1995Mar 24, 1998Matsushita Electric Industrial Co., Ltd.Scene change detecting method for video and movie
US5767923 *Jun 7, 1996Jun 16, 1998Electronic Data Systems CorporationMethod and system for detecting cuts in a video signal
US5778108 *Jun 7, 1996Jul 7, 1998Electronic Data Systems CorporationMethod and system for detecting transitional markers such as uniform fields in a video signal
US5805733 *Dec 12, 1994Sep 8, 1998Apple Computer, Inc.Method and system for detecting scenes and summarizing video sequences
US5821945 *May 15, 1997Oct 13, 1998The Trustees Of Princeton UniversityMethod and apparatus for video browsing based on content and structure
US5864682 *May 21, 1997Jan 26, 1999Oracle CorporationMethod and apparatus for frame accurate access of digital audio-visual information
US5874986 *Jun 26, 1996Feb 23, 1999At&T CorpMethod for communicating audiovisual programs over a communications network
US5920360 *Jun 7, 1996Jul 6, 1999Electronic Data Systems CorporationMethod and system for detecting fade transitions in a video signal
US5956026 *Dec 19, 1997Sep 21, 1999Sharp Laboratories Of America, Inc.Method for hierarchical summarization and browsing of digital video
US5959697 *Jun 7, 1996Sep 28, 1999Electronic Data Systems CorporationMethod and system for detecting dissolve transitions in a video signal
US5973723 *Dec 12, 1997Oct 26, 1999Deluca; Michael JosephSelective commercial detector and eliminator apparatus and method
US5990980 *May 28, 1998Nov 23, 1999Sarnoff CorporationDetection of transitions in video sequences
US5991365 *Mar 12, 1997Nov 23, 1999Siemens Corporate Research, Inc.Remote phone-based access to a universal multimedia mailbox
US5995095 *May 21, 1999Nov 30, 1999Sharp Laboratories Of America, Inc.Method for hierarchical summarization and browsing of digital video
US6008802 *Jan 5, 1998Dec 28, 1999Intel CorporationMethod and apparatus for automatically performing a function based on the reception of information corresponding to broadcast data
US6023520 *Jul 8, 1996Feb 8, 2000Hitach, Ltd.Method and apparatus for detecting and displaying a representative image of a shot of short duration in a moving image
US6026389 *Aug 4, 1997Feb 15, 2000Kokusai, Denshin, Denwa, Kabushiki KaishaVideo query and editing system
US6052492 *Dec 9, 1997Apr 18, 2000Sun Microsystems, Inc.System and method for automatically generating an image to represent a video sequence
US6061471 *Dec 18, 1997May 9, 2000Electronic Data Systems CorporationMethod and system for detecting uniform images in video signal
US6118923 *May 5, 1997Sep 12, 2000Intel CorporationMethod and apparatus for deferred selective viewing of televised programs
US6166735 *Dec 3, 1997Dec 26, 2000International Business Machines CorporationVideo story board user interface for selective downloading and displaying of desired portions of remote-stored video data objects
US6181817Nov 16, 1998Jan 30, 2001Cornell Research Foundation, Inc.Method and system for comparing data objects using joint histograms
US6222532 *Jan 29, 1998Apr 24, 2001U.S. Philips CorporationMethod and device for navigating through video matter by means of displaying a plurality of key-frames in parallel
US6243419 *May 27, 1997Jun 5, 2001Nippon Telegraph And Telephone CorporationScheme for detecting captions in coded video data without decoding coded video data
US6285788May 27, 1998Sep 4, 2001Sharp Laboratories Of America, Inc.Method for fast return of abstracted images from a digital image database
US6310625Sep 25, 1998Oct 30, 2001Matsushita Electric Industrial Co., Ltd.Clip display method and display device therefor
US6323877Oct 22, 1997Nov 27, 2001Matsushita Electric Industrial Co., Ltd.Picture display unit, picture display system, and moving picture retrieving system
US6341168Apr 6, 1999Jan 22, 2002Hitachi, Ltd.Method and apparatus for detecting and displaying a representative image of a shot of short duration in a moving image
US6342904Dec 17, 1998Jan 29, 2002Newstakes, Inc.Creating a slide presentation from full motion video
US6356658Apr 11, 2000Mar 12, 2002Sharp Laboratories Of America, Inc.Method for fast return of abstracted images from a digital image database
US6360234Aug 14, 1998Mar 19, 2002Virage, Inc.Video cataloger system with synchronized encoders
US6393054Apr 20, 1998May 21, 2002Hewlett-Packard CompanySystem and method for automatically detecting shot boundary and key frame from a compressed video data
US6445835Sep 28, 2000Sep 3, 2002Sharp Laboratories Of America, Inc.Method for image characterization using color and texture statistics with embedded spatial information
US6463444Aug 14, 1998Oct 8, 2002Virage, Inc.Video cataloger system with extensibility
US6529638Feb 1, 1999Mar 4, 2003Sharp Laboratories Of America, Inc.Block boundary artifact reduction for block-based image compression
US6542632Mar 8, 1999Apr 1, 2003Sharp Laboratories Of America, Inc.Method for image characterization using color and texture statistics with embedded spatial information
US6557003 *Mar 28, 2000Apr 29, 2003Genesys Telecommunications Laboratories, Inc.Method and apparatus for multiple agent commitment tracking and notification
US6567980Aug 14, 1998May 20, 2003Virage, Inc.Video cataloger system with hyperlinked output
US6580437 *Jun 26, 2000Jun 17, 2003Siemens Corporate Research, Inc.System for organizing videos based on closed-caption information
US6587123 *Mar 23, 1999Jul 1, 2003Sony CorporationVideo material selecting apparatus and method for selecting video material
US6714216 *Sep 27, 1999Mar 30, 2004Sony CorporationVideo editing apparatus and method
US6721952 *Mar 7, 1997Apr 13, 2004Roxio, Inc.Method and system for encoding movies, panoramas and large images for on-line interactive viewing and gazing
US6760536May 16, 2000Jul 6, 2004International Business Machines CorporationFast video playback with automatic content based variable speed
US6760917 *Mar 28, 1996Jul 6, 2004Sony Europa B.V.System for information on demand
US6784879 *Jul 14, 1997Aug 31, 2004Ati Technologies Inc.Method and apparatus for providing control of background video
US6801294 *Nov 21, 2002Oct 5, 2004Samsung Electronics Co., Ltd.Recording and/or reproducing apparatus and method using key frame
US6807362 *Jul 18, 2000Oct 19, 2004Fuji Xerox Co., Ltd.System and method for determining video clip boundaries for interactive custom video creation system
US6807367Jan 2, 2000Oct 19, 2004David DurlachDisplay system enabling dynamic specification of a movie's temporal evolution
US6845176Mar 17, 1998Jan 18, 2005Sharp Laboratories Of America, Inc.System and method for on-line computation and storage of scene histograms in an image acquisition appliance
US6877134Jul 29, 1999Apr 5, 2005Virage, Inc.Integrated data and real-time metadata capture system and method
US6892351Jan 17, 2002May 10, 2005Newstakes, Inc.Creating a multimedia presentation from full motion video using significance measures
US6925207Mar 23, 2000Aug 2, 2005Sharp Laboratories Of America, Inc.Method for fast return of abstracted images from a digital images database
US6931595Mar 30, 2001Aug 16, 2005Sharp Laboratories Of America, Inc.Method for automatic extraction of semantically significant events from video
US6957387 *Sep 5, 2001Oct 18, 2005Koninklijke Philips Electronics N.V.Apparatus for reproducing an information signal stored on a storage medium
US6964021 *Aug 17, 2001Nov 8, 2005Lg Electronics Inc.Method and apparatus for skimming video data
US6965706Jan 31, 2002Nov 15, 2005Sharp Laboratories Of America, Inc.Method for image characterization using color and texture statistics with embedded spatial information
US7006945Mar 31, 2003Feb 28, 2006Sharp Laboratories Of America, Inc.Processing of video content
US7047157Oct 30, 2004May 16, 2006Sharp Laboratories Of America, Inc.Processing of video content
US7062084Dec 4, 2000Jun 13, 2006Sharp Laboratories Of America, Inc.Method for image description using color and local spatial information
US7065250Sep 15, 1999Jun 20, 2006Canon Kabushiki KaishaAutomated image interpretation and retrieval system
US7076114Jan 3, 2003Jul 11, 2006Sharp Laboratories Of America, Inc.Block boundary artifact reduction for block-based image compression
US7089496Jul 20, 2001Aug 8, 2006Hewlett-Packard Development Company, L.P.System and method for scene detection information storage
US7092620Aug 5, 1999Aug 15, 2006Hewlett-Packard Development Company, L.P.Converting analog video data into digital form
US7093191Dec 21, 2001Aug 15, 2006Virage, Inc.Video cataloger system with synchronized encoders
US7120873Jan 28, 2002Oct 10, 2006Sharp Laboratories Of America, Inc.Summarization of sumo video content
US7131059Dec 31, 2002Oct 31, 2006Hewlett-Packard Development Company, L.P.Scalably presenting a collection of media objects
US7143353 *Mar 30, 2001Nov 28, 2006Koninklijke Philips Electronics, N.V.Streaming video bookmarks
US7143354Aug 20, 2001Nov 28, 2006Sharp Laboratories Of America, Inc.Summarization of baseball video content
US7143362Dec 28, 2001Nov 28, 2006International Business Machines CorporationSystem and method for visualizing and navigating content in a graphical user interface
US7149755Jul 29, 2002Dec 12, 2006Hewlett-Packard Development Company, Lp.Presenting a collection of media objects
US7152210 *Oct 18, 2000Dec 19, 2006Koninklijke Philips Electronics N.V.Device and method of browsing an image collection
US7167809Oct 30, 2004Jan 23, 2007Sharp Laboratories Of America, Inc.Processing of video content
US7177470Nov 13, 2002Feb 13, 2007Koninklijke Philips Electronics N. V.Method of and system for detecting uniform color segments
US7203380 *Nov 16, 2001Apr 10, 2007Fuji Xerox Co., Ltd.Video production and compaction with collage picture frame user interface
US7203620May 23, 2002Apr 10, 2007Sharp Laboratories Of America, Inc.Summarization of video content
US7295752Feb 5, 2002Nov 13, 2007Virage, Inc.Video cataloger system with audio track extraction
US7310589Oct 30, 2004Dec 18, 2007Sharp Laboratories Of America, Inc.Processing of video content
US7312812Jan 5, 2005Dec 25, 2007Sharp Laboratories Of America, Inc.Summarization of football video content
US7327891Jul 17, 2002Feb 5, 2008Yesvideo, Inc.Automatic selection of a visual image or images from a collection of visual images, based on an evaluation of the quality of the visual images
US7383504 *Mar 6, 2000Jun 3, 2008Mitsubishi Electric Research LaboratoriesMethod for representing and comparing multimedia content according to rank
US7392467 *Dec 22, 2004Jun 24, 2008Lg Electronics, Inc.Method of describing multiple level digest segment information scheme for multimedia contents and apparatus for generating digest stream from described multiple level digest segment information scheme and method thereof
US7403224Sep 10, 2004Jul 22, 2008Virage, Inc.Embedded metadata engines in digital capture devices
US7406655Dec 22, 2004Jul 29, 2008Lg Electronics, Inc.Method of describing multiple level digest segment information scheme for multimedia contents and apparatus for generating digest stream from described multiple level digest segment information scheme and method thereof
US7424677Oct 28, 2004Sep 9, 2008Sharp Laboratories Of America, Inc.Audiovisual information management system with usage preferences
US7424678Oct 28, 2004Sep 9, 2008Sharp Laboratories Of America, Inc.Audiovisual information management system with advertising
US7437674 *Oct 27, 2004Oct 14, 2008Corel Tw Corp.Video processing methods
US7440611May 12, 2005Oct 21, 2008Sharp Laboratories Of America, Inc.Method for image characterization using color and texture statistics with embedded spatial information
US7447331Feb 24, 2004Nov 4, 2008International Business Machines CorporationSystem and method for generating a viewable video index for low bandwidth applications
US7474331Jan 3, 2005Jan 6, 2009Sharp Laboratories Of America, Inc.Summarization of football video content
US7474698Sep 27, 2002Jan 6, 2009Sharp Laboratories Of America, Inc.Identification of replay segments
US7490092Feb 12, 2003Feb 10, 2009Streamsage, Inc.Method and system for indexing and searching timed media information based upon relevance intervals
US7499077Aug 20, 2001Mar 3, 2009Sharp Laboratories Of America, Inc.Summarization of football video content
US7509580Oct 28, 2004Mar 24, 2009Sharp Laboratories Of America, Inc.Audiovisual information management system with preferences descriptions
US7512698Sep 1, 2000Mar 31, 2009Broadband Royalty CorporationDynamic quality adjustment based on changing streaming constraints
US7559015 *Dec 11, 2003Jul 7, 2009Panasonic CorporationData processing device and method for selecting media segments on the basis of a score
US7594245Jun 13, 2005Sep 22, 2009Sharp Laboratories Of America, Inc.Networked video devices
US7639275Jan 3, 2005Dec 29, 2009Sharp Laboratories Of America, Inc.Summarization of football video content
US7653131Dec 2, 2005Jan 26, 2010Sharp Laboratories Of America, Inc.Identification of replay segments
US7657836Sep 27, 2002Feb 2, 2010Sharp Laboratories Of America, Inc.Summarization of soccer video content
US7657907Sep 30, 2002Feb 2, 2010Sharp Laboratories Of America, Inc.Automatic user profiling
US7676820Dec 19, 2003Mar 9, 2010Koninklijke Philips Electronics N.V.Method and apparatus for similar video content hopping
US7680343 *Feb 4, 2008Mar 16, 2010Yes Video, Inc.Automatic selection of a visual image or images from a collection of visual images, based on an evaluation of the quality of the visual images
US7725828 *Oct 15, 2003May 25, 2010Apple Inc.Application of speed effects to a video presentation
US7760908Mar 31, 2005Jul 20, 2010Honeywell International Inc.Event packaged video sequence
US7769827May 22, 2007Aug 3, 2010Virage, Inc.Interactive video application hosting
US7793205Jul 8, 2005Sep 7, 2010Sharp Laboratories Of America, Inc.Synchronization of video and data
US7802188 *May 13, 2004Sep 21, 2010Hewlett-Packard Development Company, L.P.Method and apparatus for identifying selected portions of a video stream
US7848598 *Sep 15, 2003Dec 7, 2010Fuji Xerox Co., Ltd.Image retrieval processing to obtain static image data from video data
US7853865Jul 8, 2005Dec 14, 2010Sharp Laboratories Of America, Inc.Synchronization of video and data
US7873905 *Jul 24, 2003Jan 18, 2011Fuji Xerox Co., Ltd.Image processing system
US7904814Dec 13, 2001Mar 8, 2011Sharp Laboratories Of America, Inc.System for presenting audio-video content
US7962948Apr 6, 2001Jun 14, 2011Virage, Inc.Video-enabled community building
US8010903Dec 28, 2001Aug 30, 2011International Business Machines CorporationSystem and method for visualizing and navigating dynamic content in a graphical user interface
US8018491Jan 3, 2005Sep 13, 2011Sharp Laboratories Of America, Inc.Summarization of football video content
US8020183Mar 30, 2001Sep 13, 2011Sharp Laboratories Of America, Inc.Audiovisual management system
US8028234Mar 8, 2005Sep 27, 2011Sharp Laboratories Of America, Inc.Summarization of sumo video content
US8028314May 26, 2000Sep 27, 2011Sharp Laboratories Of America, Inc.Audiovisual information management system
US8031947 *Mar 24, 2008Oct 4, 2011Jacobsen Kenneth PMethod and system for rapid matching of video streams
US8117206Jan 7, 2009Feb 14, 2012Streamsage, Inc.Method and system for indexing and searching timed media information based upon relevance intervals
US8150239Apr 16, 2003Apr 3, 2012Immotion Imagery Technologies, LLCPicture-based video indexing system
US8171509Jun 18, 2004May 1, 2012Virage, Inc.System and method for applying a database to video multimedia
US8176043Mar 12, 2009May 8, 2012Comcast Interactive Media, LlcRanking search results
US8176425 *Jan 31, 2002May 8, 2012Ensequence, Inc.Animated screen object for annotation and selection of video sequences
US8196043Nov 5, 2007Jun 5, 2012Apple Inc.User interface for presenting media information
US8209612Apr 19, 2010Jun 26, 2012Apple Inc.Application of speed effects to a video presentation
US8214741May 22, 2002Jul 3, 2012Sharp Laboratories Of America, Inc.Synchronization of video and data
US8314789Aug 28, 2008Nov 20, 2012Autodesk, Inc.Navigation system for a 3D virtual scene
US8356317Jun 13, 2005Jan 15, 2013Sharp Laboratories Of America, Inc.Presence based technology
US8359537Jan 15, 2010Jan 22, 2013Apple Inc.Tool for navigating a composite presentation
US8387087Apr 27, 2012Feb 26, 2013Virage, Inc.System and method for applying a database to video multimedia
US8392834Apr 9, 2003Mar 5, 2013Hewlett-Packard Development Company, L.P.Systems and methods of authoring a multimedia file
US8423555Jul 9, 2010Apr 16, 2013Comcast Cable Communications, LlcAutomatic segmentation of video
US8495694May 12, 2011Jul 23, 2013Virage, Inc.Video-enabled community building
US8527520Jan 11, 2012Sep 3, 2013Streamsage, Inc.Method and system for indexing and searching timed media information based upon relevant intervals
US8533223May 12, 2009Sep 10, 2013Comcast Interactive Media, LLC.Disambiguation and tagging of entities
US8548978Aug 21, 2007Oct 1, 2013Virage, Inc.Network video guide and spidering
US8555170Aug 10, 2010Oct 8, 2013Apple Inc.Tool for presenting and editing a storyboard representation of a composite presentation
US8606782Jun 14, 2004Dec 10, 2013Sharp Laboratories Of America, Inc.Segmentation description scheme for audio-visual content
US8612858May 1, 2009Dec 17, 2013Apple Inc.Condensing graphical representations of media clips in a composite display area of a media-editing application
US8621357 *Jul 2, 2009Dec 31, 2013Apple Inc.Light table for editing digital media
US8626322Jul 2, 2009Jan 7, 2014Apple Inc.Multimedia display based on audio and visual complexity
US8665272Aug 28, 2008Mar 4, 2014Autodesk, Inc.Navigation system for a 3D virtual scene
US8686991Aug 28, 2008Apr 1, 2014Autodesk, Inc.Navigation system for a 3D virtual scene
US8689253Mar 3, 2006Apr 1, 2014Sharp Laboratories Of America, Inc.Method and system for configuring media-playing sets
US8706735 *Jul 31, 2013Apr 22, 2014Streamsage, Inc.Method and system for indexing and searching timed media information based upon relevance intervals
US8713016Dec 24, 2008Apr 29, 2014Comcast Interactive Media, LlcMethod and apparatus for organizing segments of media assets and determining relevance of segments to a query
US8744249Jun 17, 2011Jun 3, 2014Apple Inc.Picture selection for video skimming
US8745499May 25, 2011Jun 3, 2014Apple Inc.Timeline search and index
US8749544Aug 28, 2008Jun 10, 2014Autodesk, Inc.Navigation system for a 3D virtual scene
US8776142Sep 2, 2009Jul 8, 2014Sharp Laboratories Of America, Inc.Networked video devices
US8803881Aug 28, 2008Aug 12, 2014Autodesk, Inc.Navigation system for a 3D virtual scene
US8819557Jul 15, 2010Aug 26, 2014Apple Inc.Media-editing application with a free-form space for organizing or compositing media clips
US8832555 *Jul 2, 2009Sep 9, 2014Apple Inc.Framework for slideshow object
US20100169777 *Jul 2, 2009Jul 1, 2010Apple Inc.Light Table for Editing Digital Media
US20100169783 *Jul 2, 2009Jul 1, 2010Apple, Inc.Framework for Slideshow Object
US20100303365 *May 29, 2009Dec 2, 2010Min ZhangMethods and apparatus to monitor a multimedia presentation including multiple content windows
US20110167390 *Mar 9, 2011Jul 7, 2011Ingram Dv LlcApparatus and method for utilizing an information unit to provide navigation features on a device
US20130318121 *Jul 31, 2013Nov 28, 2013Streamsage, Inc.Method and System for Indexing and Searching Timed Media Information Based Upon Relevance Intervals
USH1684 *Sep 29, 1995Oct 7, 1997Xerox CorporationFast preview processing for JPEG compressed images
USRE42905Mar 28, 1996Nov 8, 2011Sony Europa B.V.System for information on demand
CN1133128C *Oct 22, 1997Dec 31, 2003松下电器产业株式会社Picture display unit, picture display system, and moving picture retrieving system
CN1735887BDec 19, 2003Apr 28, 2010皇家飞利浦电子股份有限公Method and apparatus for similar video content hopping
CN100392749CSep 5, 2001Jun 4, 2008皇家菲利浦电子有限公司Apparatus for reproducing information signal stored on storage medium
EP0838766A2 *Oct 21, 1997Apr 29, 1998Matsushita Electric Industrial Co., Ltd.Picture Display unit, picture display system, and moving picture retrieving system
EP0907147A2 *Sep 25, 1998Apr 7, 1999Matsushita Electric Industrial Co., Ltd.Clip display method and display device therefor
EP1180253A1 *Apr 17, 2000Feb 20, 2002Apple Computer, Inc.User interface for presenting media information
EP1460566A2 *Oct 21, 1997Sep 22, 2004Matsushita Electric Industrial Co., Ltd.Picture display unit, picture display system, and moving picture retrieving system
EP2256588A1 *Apr 17, 2000Dec 1, 2010Apple Inc.User interface for presenting media information
EP2256589A1 *Apr 17, 2000Dec 1, 2010Apple Inc.User interface for presenting media information
WO1999033261A1 *Dec 22, 1998Jul 1, 1999Sarnoff CorpDetection of transitions in video sequences
WO2002005135A2Jun 29, 2001Jan 17, 2002Streamsage IncMethod and system for indexing and searching timed media information based upon relevance intervals
WO2002021529A1 *Sep 5, 2001Mar 14, 2002Mauro BarbieriAn apparatus for reproducing an information signal stored on a storage medium
WO2002084667A1 *Apr 9, 2002Oct 24, 2002Koninkl Philips Electronics NvMethod and system for selecting a position in an image sequence
WO2003009216A1 *Jul 17, 2002Jan 30, 2003Yesvideo IncAutomatic selection of a visual image based on quality
WO2004061711A1 *Dec 19, 2003Jul 22, 2004Koninkl Philips Electronics NvMethod and apparatus for similar video content hopping
Classifications
U.S. Classification715/723, G9B/27.012, 348/E05.067, G9B/27.029, 345/589, 707/E17.028, G9B/27.051, 715/716
International ClassificationG11B27/034, H04N5/14, G11B27/34, G06F17/30, G06F3/048, G09B5/06, G06F3/033, G11B27/28
Cooperative ClassificationG11B27/34, G11B27/28, G09B5/065, H04N5/147, G11B27/034, G06F3/0485, G06F17/30802, G06F17/30805, G06F17/30852
European ClassificationG06F17/30V1V2, G06F17/30V1V1, G06F17/30V5D, G06F3/0485, H04N5/14S, G11B27/34, G11B27/034, G09B5/06C, G11B27/28
Legal Events
DateCodeEventDescription
Apr 5, 2010ASAssignment
Owner name: SIEMENS CORPORATION,NEW JERSEY
Free format text: MERGER;ASSIGNOR:SIEMENS CORPORATE RESEARCH, INC.;REEL/FRAME:024185/0042
Effective date: 20090902
Oct 16, 2007FPAYFee payment
Year of fee payment: 12
Oct 8, 2003FPAYFee payment
Year of fee payment: 8
Oct 18, 1999FPAYFee payment
Year of fee payment: 4
Jul 15, 1994ASAssignment
Owner name: SIEMENS CORPORATE RESEARCH, INC., NEW JERSEY
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:DEPOMMIER, REMI;REEL/FRAME:007070/0174
Effective date: 19940618
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:ARMAN, FARSHID;HSU, ARDING;CHIU, MING-YEE;REEL/FRAME:007070/0430
Effective date: 19940526